Characteristics of the Included Studies
We followed the PRISMA guidelines for literature retrieval, and the detailed description was shown in Figure 1. A total of 79 studies, involving 43 NIBS studies [17, 21, 28, 52-56] [20, 23, 24, 50, 57-82],and 36 PE studies [34, 37, 49, 83-115]. met the inclusion criteria. Figure 2 and (Supplementary file1, Figure S1) showed the degree of bias risk of the included studies. The risks in randomness, outcome data integrity and the results were low in most studies. In contrast, allocation of concealment patterns and blindness of patients and researchers were often unclear. The overall quality included in our study was moderate.
The protocol for the present review was registered to PROSPERO (registration number: CRD42022380500).
[FIGURE 1 ABOUT HERE]
Figure 1. Screening flow chart.
[FIGURE 2 ABOUT HERE]
Figure 2. Risk of bias graph.
NIBS: A total of 43 studies with 1298 patients were identified, including 23 TMS studies, 19 tES studies, and one tRNS studies. A total of 30 studies were parallel designs and others used crossover design. A total of 25 RCTs included patients with AD, 16 RCTs included patients with MCI, and two studies included patients with MCI and early-stage AD. Demographic information (e.g., age, gender etc.), intervention parameters, evaluation results, adverse reports, and follow-up information of these studies were described in Table 1.
[TABLE 1 ABOUT HERE]
PE: A total of 36 studies with 3002 patients were identified, including 35 studies were parallel design and one crossover design. A total of 10 RCTs included patients with AD, 24 RCT included patients with MCI, and two studies included patients with MCI and mild AD. Demographic information (e.g., age, gender), intervention parameters, evaluation results, adverse reports, and follow-up information of these studies were described in Table 2.
[TABLE 2 ABOUT HERE]
Global Cognition
NIBS: A total of 19 studies with 735 AD patients and six studies with 155 MCI patients reported the effect on the global cognition. The results showed that NIBS had no significant effect on the global cognition in AD (SMD=0.11 p=0.53) or MCI (SMD=0.56, p=0.08) (see Table 3 and Supplementary file1, Figure S2). The corresponding funnel plot was approximately symmetrical, indicating limited publication bias (see Supplementary file1, Figure S3).
PE: A total of 11 studies with 596 AD patients and 23 studies with 1769 MCI patients reported the effect on the global cognition. The results showed that PE had significant effect on global cognition in AD (SMD=0.57, p=0.004) and MCI (SMD=0.8, p<0.00001) (see Table 3 and Supplementary file1, Figure S4). The corresponding funnel plot was approximately symmetrical which indicated no significant publication bias (see Supplementary file1, Figure S5).
Memory
NIBS: A total of 16 studies with 599 AD patients and eight studies with 340 MCI patients reported the effect on memory. The results showed that NIBS had significant effect on memory in AD (SMD=0.7, p=0.0002) and MCI (SMD=0.71, p<0.0001) (see Table 3 and Supplementary file1, Figure S6). The corresponding funnel plot was approximately symmetrical which indicated no significant publication bias (see Supplementary file1, Figure S7).
PE: Three studies with 1300 AD patients and eight studies with 1129 MCI patients reported the effect on memory. The results showed that PE had significant effect on memory in AD (SMD=0.61, p=0.02), but not in MCI (SMD=0.21, p=0.05) (see Table 3 and Supplementary file1, Figure S8). The corresponding funnel plots was approximately symmetrical which indicated no significant publication bias (see Table 3 and Supplementary file1, Figure S9).
Executive Function
NIBS: Six studies with 181 AD patients and seven studies with 180 MCI patients reported the effect on the executive function. The results showed that NIBS had significant effect on executive function in AD (SMD=0.39, p=0.01), but not in MCI (SMD=0.24, p=0.12) (see Table 3 and Supplementary file1, Figure S10). The corresponding funnel diagram was not significantly symmetrical, indicating a slight publication bias (see Supplementary file1, Figure S11).
PE: Five studies with 568 AD patients and 13 studies with 981 MCI patients reported the effect on the executive function. The results showed that PE had significant effect on executive function in AD (SMD=0.22, p=0.14), or MCI (SMD=0.27, p=0.13) (see Table 3 and Supplementary file1, Figure S12). The corresponding funnel plots was approximately symmetrical which indicated no significant publication bias (see Supplementary file1, Figure S13).
Language Function
NIBS: Seven studies with 273 AD patients and three studies with 116 MCI patients reported the effect on the language function. The results showed that NIBS had significant effect on the language function in AD (SMD=1.08, p=0.0005), but not in MCI (SMD=0.61, p=0.08) (see Table 3 and Supplementary file1, Figure S14). The corresponding funnel plots was approximately symmetrical which indicated no significant publication bias (see Supplementary file1, Figure S15).
PE: Six studies with 585 AD patients and seven studies with 807 MCI patients reported the effect on the language function. The results showed that PE had significant effect on the language function in AD (SMD=0.64, p=0.02), but not in MCI (SMD=0.08, p=0.41) (see Table 3 and Supplementary file1, Figure S16). The corresponding funnel diagram was not significantly symmetrical, indicating a slight publication bias (see Supplementary file1, Figure S17).
Attention
NIBS: Four studies with 174 AD patients and five studies with 114 MCI patients reported the effect on attention. The results showed that NIBS had significant effect on attention in AD (SMD=0.55, p=0.0004), but not in MCI (SMD=0.28, p=0.28) (see Table 3 and Supplementary file1, Figure S18). The corresponding funnel plot was approximately symmetrical which indicated very small publication bias (see Supplementary file1, Figure S19).
PE: Two studies with 210 AD patients and five studies with 546 MCI patients reported the effect on attention. The results showed that PE had no significant effect on attention in AD (SMD=0.06, p=0.65) or MCI (SMD=0.25, p=0.08) (see Table 3 and Supplementary file1, Figure S20). The corresponding funnel plot was approximately symmetrical which indicated very small publication bias (see Supplementary file1, Figure S21).
Long-term Effect on Global Cognition
As for the long-term effect of the two interventions on AD or MCI, only the scale of global cognitive function was analyzed, due to the limited number of studies and the lack of neurocognitive scales.
NIBS: Nine studies with 367 patients reported the long-term effects of NIBS on the global cognitive function in patients with AD or MCI. The results showed that NIBS had significant long-term sustained effect on the global cognitive function in AD and MCI (SMD=0.52, p=0.01) (see Table 3 and Supplementary file1, Figure S22). The corresponding funnel plot was approximately symmetrical which indicated no significant publication bias (see Supplementary file1, Figure S23).
PE: Five studies with 223 patients reported the long-term effects of PE on the global cognitive function in patients with AD or MCI. The results showed that PE had significant long-term sustained effect on the global cognitive function in AD or MCI (SMD=0.94, p=0.005) (see Table 3 and Supplementary file1, Figure S24). The corresponding funnel plot was approximately symmetrical which indicated no significant publication bias (see Supplementary file1, Figure S25).
[TABLE 3 ABOUT HERE]
Sensitivity Analyses
One sensitivity analysis was conducted to determine if removing these studies would significantly affect pooled effect estimates. The analysis of the effect of PE on memory in individuals with MCI showed that the result after excluding one study was significant with heterogeneity < 50% (SMD=0.26, p=0.01) (see Supplementary file1, Figure S26).
Adverse Events
NIBS: A total of 12 studies including a total of 61 patients reported adverse events, including headache, cervical pain, scalp pain, pruritus, dizziness, shock, and facial convulsion, etc. These adverse effects were mild and transient which would disappear without any special treatment.
PE: Two studies including a total of 62 adverse events were reported, including musculoskeletal problems, dizziness or faintness, symptoms related to AD, somatic disease (i.e., cold, anemia, erysipelas), pneumonia, urinary tract infection, and pulmonary edema, etc. The most common adverse events were musculoskeletal problems, and only a few of them were directly related to PE intervention.